the impacts of water infrastructure and climate change on ......infrastructure and climate change on...

Luna Bharati, Guillaume Lacombe, Pabitra Gurung, Priyantha Jayakody, Chu Thai Hoanh and Vladimir Smakhtin

The Impacts of Water Infrastructure and Climate Change on the Hydrology of the Upper Ganges River Basin

142

RESEARCHIWMI

R E P O R T

I n t e r n a t i o n a lWater ManagementI n s t i t u t e

Research Reports

The publications in this series cover a wide range of subjects—from computer

modeling to experience with water user associations—and vary in content from

directly applicable research to more basic studies, on which applied work ultimately

depends. Some research reports are narrowly focused, analytical and detailed

empirical studies; others are wide-ranging and synthetic overviews of generic

problems.

Although most of the reports are published by IWMI staff and their

collaborators, we welcome contributions from others. Each report is reviewed

internally by IWMI staff, and by external reviewers. The reports are published and

distributed both in hard copy and electronically (www.iwmi.org) and where possible

��

may be copied freely and cited with due acknowledgment.

About IWMI

IWMI’s mission is to improve the management of land and water resources for food, livelihoods and the environment. In serving this mission, IWMI concentrates

on the integration of policies, technologies and management systems to achieve

��

irrigation and water and land resources.

i

International Water Management Institute

P O Box 2075, Colombo, Sri Lanka

IWMI Research Report 142

The Impacts of Water Infrastructure and

Climate Change on the Hydrology of the

Upper Ganges River Basin


The authors: Luna Bharati is a Senior Researcher and Head of the Nepal office of the International

Water Management Institute (IWMI) in Kathmandu, Nepal ([email protected]); Guillaume Lacombe is

a Researcher based at the Laos office of IWMI in Vientiane, Lao PDR ([email protected]); Pabitra

Gurung is a Research Officer at the Nepal office of IWMI in Kathmandu, Nepal ([email protected]);

Priyantha Jayakody was a Research Officer based at the headquarters of IWMI in Colombo, Sri Lanka,

and is now a PhD student at the Mississippi State University, Agriculture and Biological Department in

Starkville, Mississippi, United States ([email protected]); Chu Thai Hoanh is a Principal Researcher

and Acting Head of the Laos office of IWMI in Vientiane, Lao PDR ([email protected]); and Vladimir

Smakhtin is Theme Leader - Water Availability and Access at the headquarters of IWMI in Colombo, Sri

Lanka ([email protected]).

Bharati, L.; Lacombe, G.; Gurung, P.; Jayakody, P.; Hoanh, C. T.; Smakhtin, V. 2011. The impacts of water infrastructure and climate change on the hydrology of the Upper Ganges River Basin. Colombo, Sri Lanka:

International Water Management Institute. 36p. (IWMI Research Report 142). doi: 10.5337/2011.210

/ water resources / river basins / climate change / hydrology / simulation models / precipitation / evapotranspiration / statistical methods / water balance / water yield / irrigation water / India / Ganges River /

ISSN 1026-0862

ISBN 978-92-9090-744-2

Copyright © 2011, by IWMI. All rights reserved. IWMI encourages the use of its material provided that the

organization is acknowledged and kept informed in all such instances.

Cover picture shows Ganga River upstream of Rishikesh, Uttarakhand, India (Photo credit: Vladimir

Smakhtin, IWMI)

Please send inquiries and comments to: [email protected]

A free copy of this publication can be downloaded at

www.iwmi.org/Publications/IWMI_Research_Reports/index.aspx

Acknowledgements

The authors would like to thank the Indian Institute of Tropical Meteorology (IITM) for providing regional

climate change projection data (Providing REgional Climates for Impacts Studies (PRECIS)) for our analysis.

We would also like to thank Nitin Kaushal, Senior Manager, World Wide Fund for Nature (WWF)-India, for

helping us with data acquisition and WWF-India for funding this study.

Project

This research study was initiated as part of WWF-India’s ‘Living Ganga Programme’.

Collaborators

The following organizations collaborated in the research conducted for this report.

International Water Management Institute (IWMI)

World Wide Fund for Nature (WWF)-India

Donors

Funding for this research was provided by the following organization.

World Wide Fund for Nature (WWF)-India

Water ManagementI n t e r n a t i o n a l

I n s t i t u t e

v

v

Contents

Summary vii

Introduction 1

Methods and Data 4

Results and Discussion 14

Conclusions 25

References 27

vii

and mountainous upper areas of the UGB. On

an annual average, present-day flows throughout

the UGB are about 2-8% lower than in naturalized

conditions. The percentage of flow reduction is the

highest during the dry months as water is being

withdrawn for irrigation. Dry and wet season flows

under CC scenario A2 (scenario corresponding

to high population growth with slower per capita

economic growth and technological change) are

lower than those in present climate conditions

at upstream locations, but higher at downstream

locations of the UGB. Flows under CC scenario B2

(corresponding to moderate population growth and

economic development with less rapid and more

diverse technological change) are systematically

higher and lower than those under CC scenario

A2 during dry and wet seasons, respectively. The

dates of minimum daily discharges are highly

variable among stations and between different CC

scenarios, while the dates of maximum flow are

delayed downstream as a result of the delay in

the onset of the monsoon in the lower parts of the

basin. The report also provides actual simulated

discharge time series data for all simulated

scenarios, in the overall attempt to augment the

river flow data for this important river basin and to

facilitate the use of these data by any interested

party.

Summary

Provision of detailed continuous long-term

hydrological time series data for any river basin is

critical for estimation, planning and management

of its current and future water resources. Most of

the river basins in India are data poor, including

its iconic river – Ganga (Ganges). This study

assessed the variability of flows under present

and ‘naturalized’ basin conditions in the Upper

Ganges Basin (UGB) (area of over 87,000 square

kilometers (km2)). The naturalized basin conditions

are those that existed prior to the development of

multiple water regulation structures, and hence

may be seen as a reference condition, a starting

point, against which to evaluate the impacts

of planned basin development, as well as the

impacts of future climate change (CC) on basin

water resources. The later impacts are also part

of the study: the PRECIS regional climate model

(RCM) was used to generate climate projections

for the UGB, with subsequent simulations of

future river flows. Results show that the annual

average precipitation, actual evapotranspiration

(ET) and net water yields of the whole basin

were 1,192 millimeters (mm), 416 mm and 615

mm, respectively. However, there were large

variations in both temporal and spatial distribution

of these components. Precipitation, ET and water

yields were found to be higher in the forested

1

The Impacts of Water Infrastructure and Climate Change on the Hydrology of the Upper Ganges River Basin


Introduction

The Ganges River System originates in the

Central Himalayas, and extends into the alluvial

Gangetic Plains and drains into the Indian Ocean

at the Bay of Bengal. Its basin area (1.09 million

km2) spreads across India (79%), Nepal (13%),

Bangladesh (4%) and China (4%). The river is

of high importance to riparian countries with an

estimated 410 million people directly or indirectly

depending on it (Verghese and Iyer 1993). In

the upstream mountainous regions, hydropower

is the main focus of development with mega

and micro projects either under construction or

being planned in both Nepal and India. After

the main river channel reaches the plains, it

is highly regulated with dams, barrages and

associated irrigation canals. All this infrastructure

development and abstractions affects the river’s

flow regime and reduces flows, which, in turn,

impacts downstream water availability, water

quality and riverine ecosystems. Furthermore,

there are concerns that CC is likely to exacerbate

the water scarcity problem in the Ganges Basin.

Therefore, modeling the hydrology of the basin is

critical for estimation, planning and management

of current and future water resources.

To operate a hydrological model, reliable data

on climatological variables such as temperature,

precipitation, evaporation, etc., over space and

time are necessary. For analysis of the past, such

information can be derived from observational

data sets. However, for assessment of the future

with the possible impacts of CC, the hydrological

models can be driven with the output from the

general circulation model (GCM) (IPCC 1996;

Akhtar et al. 2009). However, the resolutions of

GCMs are currently constrained by computational

and physical reasons to 200 kilometers (km)

for climate change predictions and are too

course for hydrological modeling at basin scale.

In order to increase the spatial resolution of

these predictions, one method that is used is

statistical downscaling techniques, which have

been developed in the last decades (e.g., Wilby

et al. 1999; Bergström et al. 2001; Pilling and

Jones 2002; Guo et al. 2002; Arnell 2003; Booij

2005; Benestad et al. 2008). A second option

is the use of dynamical downscaling (e.g., Hay

et al. 2002; Hay and Clark 2003; Fowler and

Kilsby 2007; Leander and Buishand 2007).

Dynamical downscaling fits output from GCMs

into regional meteorological models. It involves

using numerical meteorological modeling to

reflect how global patterns affect local weather

conditions. The high horizontal resolution of a

RCM (about 10-50 km) is more appropriate for

resolving the small-scale features of topography

and land use. Furthermore, the high resolution

of RCM is ideal to capture the spatial variability

of precipitation as input into hydrological models

(Gutowski et al. 2003; Akhtar et al. 2009),

and provide better representation of mountain

areas affected by the amount of rainfall and the

location of windward rainy areas and downwind

rain shadow areas (Jones et al. 2004).

2

Hydrological simulations using RCM output in

data-sparse basins such as the Ganges involves

several problems, including uncertainties in inputs,

model parameters and model structure (Akhtar

et al. 2009). The main disadvantages of RCMs

are that they inherit the large-scale errors of their

driving GCM model and require large amounts of

boundary data previously archived from relevant

GCM experiments. Additional uncertainties can be

linked to the local scale patterns in downscaling of

temperature, precipitation and evapotranspiration

in a specific basin (Bergström et al. 2001; Guo et

al. 2002; Akhtar et al. 2009).

Although modeling studies that analyzed

the impacts of water infrastructure development

on the hydrology and water resources in other

parts of the world are available, studies focusing

on the Ganges Basin are limited. The National

Communication (NATCOM) project by the Ministry

of Environment and Forests, Government of

India, quantified the impact of CC on water

resources of all major Indian river systems

(Gosain et al. 2006). This study used the Hadley

Centre Regional Climate Model 2 (HadRM2) daily

weather data as input to run the Soil and Water

Assessment Tool (SWAT) hydrological model to

determine the spatio-temporal water availability

in the river systems and to calculate basin water

balances. This study suggests that precipitation,

evapotranspiration and runoff will increase by

approximately 10% in the Ganges Basin. The

NATCOM study, however, did not consider the

effect of water infrastructure development in the

basin and modeled the Ganges without water

abstraction and use. Furthermore, the simulations

were not validated against observed flow data,

making the model results uncertain.

The Water Evaluation And Planning (WEAP)

model was used by de Condappa (2009)

for large-scale assessment of surface water

resources in the Indus and Ganges basins,

with a special focus on the contribution of snow

and glaciers to flow. Several, relatively simple

scenarios of changes in glaciated area were

simulated. Results suggest that glaciers play the

role of buffers against interannual variability in

precipitation, in particular, during years with a

weak monsoon. However, the impacts of CC and

water use in the basin were not fully assessed.

Some other modeling studies examined, in detail,

the hydrological regime of individual glaciers

in the UGB (Singh et al. 2008), rather than the

impacts of water use and CC on basin-wide water

resources. Similarly, Seidel et al. (2000) modeled

the runoff regime of the Ganges and Brahmaputra

basins, accounting for precipitation, remotely

sensed snow covered areas and temperatures

using the Snowmelt Runoff Model (SRM). They

found that the already high risk of floods during

the period July to September is slightly increased

with CC. Numerous papers can be found on

the impact of water resources development and

CC on downstream areas of Bangladesh (e.g.,

Ahmad et al. 2001; Jian et al. 2009; Mirza 2004;

Rahaman 2009). These studies, however, do not

assess the dynamics of water availability and use

in upstream areas within Nepal and India.

Apart from the general paucity of hydrological

and water resources studies with fine spatial

resolution in the Ganges Basin, the inherent

problem of this basin is the availabil ity of

observed discharge data, against which models

can be calibrated and validated. Discharge data in

the Himalayan part of the basin are scarce due to

lack of measurement stations. In the downstream

plains, although discharge data from gauging

stations exist, these data are not accessible to

the public due to national security laws in India.

This leaves most of the hydrology studies of the

Ganges, which are carried out by the government

agencies, being classified and not accessible in

the public domain. In addition, simulated data are

also not widely shared, hence impeding their use

in subsequent water resource applications.

The UGB (upstream of Kanpur Barrage;

ca tchment a rea 87 ,787 km2) (F igure 1)

encompasses various physiographic conditions,

is of great cultural and spiritual importance for

the country, yet, it is already highly regulated with

multiple water structures (with more plans on the

way), and will most likely experience significant

changes in hydrology (with significant economic

and social implications) due to CC.

This study is part of WWF-India’s “Living

Ganga Programme.” The main objective of this

programme is to develop and promote approaches

3

for sustainable water resources management,

including environmental flows which conserve

biodiversity and support livelihoods under present

and changing climate scenarios (WWF 2011).

In the present study, the flow variability of the

catchment is characterized for four scenarios

corresponding to pairwise combinations of present

and future water infrastructure development and

climate conditions, as detailed in the section,

Methods and Data. Results are presented in two

parts: 1) water balances for subcatchments for

present and ‘naturalized’ conditions (prior to the

development of multiple regulation structures), and

2) several indicators of hydrological variability that

characterize the likely impacts of CC at both present

and ‘naturalized’ flow regimes. For purposes of this

study, naturalized flows are defined as ‘free flowing

flows in the mainstream without dams and barrages,

and irrigation diversions’. These assessments

are the prerequisites for the subsequent detailed

water allocation modeling under present and future

flow regimes in the UGB. In addition, the aim of

the simulation modeling here is to provide freely

available flow time series for the UGB, in order to

enhance data availability and initiate hydrological

data sharing. All simulated data are also available

via IWMI’s water data portal (http://waterdata.iwmi.

org/).

FIGURE 1. Upper Ganges Basin (UGB) with locations of barrages, reservoirs, hydrometeorological stations and ‘environmental

flow (EF) sites’, where environmental flow assessment was undertaken under the “Living Ganga Programme.”

4

Methods and Data

The analytical framework of the study is presented

in Figure 2 and detailed in the following sections.

The SWAT hydrological model is used to simulate

flows.

The steps of the analytical framework are

detailed below.

1. SWAT model setup for the UGB: a model

with 21 subbasins was setup using a Digital

Elevation Model (DEM), soil and land use/land

cover (LULC) maps and flow data.

2. Collection of observed climate data: daily

data for five climate variables (precipitation,

maximum and minimum temperature, solar

radiation, wind speed and relative humidity)

measured inside and around the basin from

1971 to 2005 were collected.

3. Calibration and validation of SWAT model against observed flow data over the period

2000-2005.

4. Selection of a RCM for simulation of future cl imate condit ions: output t ime series

simulated by PRECIS over the periods 1961-

1990 and 2071-2100 were provided by IITM.

5. Adjus tment o f c l ima te change da ta : PRECIS t ime ser ies data for the gr id

cells encompassing climatic stations were

Analysis of Model Results and Reporting

Running Model for Different Scenarios

(i) Present Scenario without CC (1971 - 2005)

(ii) Naturalized Scenario without CC (1971 - 2005)

(iii) Present Scenario with CC-A2 (2071 - 2100)

(iv) Present Scenario with CC-B2 (2071 - 2100)

Reset ModelAdjustment of CC Data

(Observed versus Baseline)

Regional PRECIS CC DataBL [1961 - 1990]

A2 [2071 - 2100]

B2 [2071 - 2100]

Model Calibration and Validation(2000 - 2005)

Model Setup

Spatial Maps (DEM, Soil and LULC)

Observed Climate Data (1971 - 2005) Precipitation

Temperature

Relative humidity

Wind speed

Solar radiation

Observed Flow Data

FIGURE 2. Analytical framework. Note: BL = baseline scenario; A2 = Climate conditions as projected by PRECIS under A2

scenarios; B2 = Climate conditions as projected by PRECIS under B2 scenarios

5

compared to actual records over the period

1961-1990 to statistically determine the

required bias adjustments. These adjustments

were then applied to the projection period

2071-2100 based on the assumption that

the same bias occurs in both simulation and

projection periods.

6. Setting study scenarios: based on the

objectives of the study, four scenarios with

different land and water use, and climate

conditions were established. These are

( i ) present condi t ion scenar io (water

infrastructure development as of 2005 and

present climate as measured from 1971 to

2005); (ii) ‘naturalized’ scenario, assuming

that no water infrastructure were built under

present climate conditions (1971-2005); (iii)

climate change scenarios, assuming that

water infrastructure is that of 2005, under

2071-2100 climate conditions as projected

by PRECIS under A2 scenarios; and (iv) B2

scenarios.

7. Scenario simulation using SWAT model.

8. Analysis of simulated results: simulated

water balances were compared between

present and ‘naturalized’ conditions. The

impacts of c l imate change and water

infrastructure developments on flow regimes

at the four ‘EF sites’ were characterized

using indicators of hydrological alterations.

This report presents the main results of this

analytical framework.

SWAT Model Description and Setup

SWAT is a process-based continuous hydrological

mode l tha t p red ic ts the impac t o f land

management practices on water, sediment and

agricultural chemical yields in complex basins

with varying soils, land use and management

conditions (Arnold et al. 1998; Srinivasan et al.

1998). The main components of the model include

climate, hydrology, erosion, soil temperature, plant

growth, nutrients, pesticides, land management,

channel and reservoir routing. SWAT divides a

basin into subbasins. Subbasins are connected

through a stream channel. Each subbasin is

further divided into Hydrological Response Units

(HRU). A HRU is a unique combination of soil

and vegetation types. SWAT simulates hydrology,

vegetation growth and management practices at

the HRU level.

The hydrological cycle is simulated by SWAT

using the following water balance equation (1):

)1(SWt = SWo + (Rday - Qsurf - Ea - wseep - Qgw )Σ

n

i = 1

where: SWt: Final soil water content (mm); SWo: Initial soil water content (mm); t: Time in days; Rday: pre-

cipitation on day i (mm); Qsurf : surface runoff on day i (mm); Ea: actual evapotranspiration on day i (mm);

wseep: percolation on day i (mm); and Qgw: return flow on day i (mm).

Basin subdivision allows differences in

evapotranspiration for various crops and soils

to be simulated. Runoff is predicted separately

for each subbasin and routed to obtain the total

runoff for the basin. This increases the accuracy

and gives a much better physical description of

the water balance. More detailed descriptions of

the model can be found in Arnold et al. (1998),

Srinivasan et al. (1998) and Neitsch et al. 2005.

The SWAT model was chosen for this study

because it can be used in large agricultural river

basin scales to also simulate crop water use.

As actual data for irrigation distribution was not

available, the calculated crop water use helped

determine the irrigation water requirements in the

basin. The SWAT model has been extensively

used in the USA and elsewhere for calculating

water balances of primarily agricultural catchments

(e.g., Jha 2011; Bharati and Jayakody 2011; Garg

et al. 2011).

6

The SWAT model requires three basic files for

delineating the basin into subbasins and HRUs:

Digital Elevation Model (DEM), soil map and Land

Use/Land Cover (LULC) map. Figure 3(a) shows

the DEM for the basin using 90 m Shuttle Radar

Topography Mission (SRTM) data.

The elevation in the UGB ranges from 100 m

in the lower plains to 7,500 m in the upper

mountain region. Some mountain peaks in the

headwater basin are permanently covered by

snow. Figure 3(b) shows the land use map which

was developed using the LandSat image from

2003. According to this map, around 65% of

the basin is occupied by agriculture. The main

crop types (identified from the land use map

and district statistics) are wheat, maize, rice,

sugarcane, pearl millet and potato. About 25%

of the area of the UGB is covered by forests,

mostly in the upper mountains. For the naturalized

scenarios, as there are no water provisions for

irrigation, the irrigated crops such as rice and

sugarcane are replaced by rainfed crops from the

region, i.e., lentils and wheat.

Figure 3(c) shows the soil map of the basin.

There are nine soil types in the UGB. Lithosols

dominate the upper, steep mountainous areas

and are very shallow and erodible. Cambisols and

luvisols are found in the lower areas. Cambisols

are developed in medium- and fine-textured

material derived from alluvial, colluvial and

aeolian deposits. Most of these soils make good

agricultural land. Luvisols are tropical soils mostly

used by farmers because of its ease of cultivation,

but they are greatly affected by water erosion and

loss in fertility. Annual average precipitation in the

UGB ranges from 550 to 2,500 mm (Figure 3(d)). A

major part of the rainfall is due to the southwestern

monsoon from July to October. Wet season

corresponds to the period July to November and

the dry season extends from December to June.

The UGB mainstream is highly regulated with

dams, barrages and corresponding canal systems

(Figure 1). The two main dams, Ramganga and

Tehri, became operational in 1988 and 2008, and

have total storage capacities of 2,448 and 3,540

106 m

3, respectively. There are three main canal

systems. The Upper Ganga canal takes off from

the right flank of the Bhimgoda barrage with a

head discharge of 190 m3/s, and irrigates an area

of 2 million hectares (Mha). The Madhya Ganga

canal provides annual irrigation water to 178,000

hectares (ha). The Lower Ganga canal from the

Narora weir irrigates 0.5 Mha.

During the winter season, a part of the runoff

in the basin is generated through contributions

from snowmelt and glacier melt. Therefore, for

this study, snowmelt was computed using the

SWAT model. In the model, when the mean

daily air temperature is less than the snowfall

temperature, the precipitation within a HRU

is classified as snow and the l iquid water

equivalent of the snow precipitation is added

to the snowpack. The snowpack increases with

additional snowfall, but decreases with snowmelt

or sublimation. In the model, snowmelt is

controlled by the air and snowpack temperature,

the melting rate and the areal coverage of snow.

Snowmelt is then included with rainfall in the

calculations of runoff and percolation. Further

information on snowmelt calculations can be

found in Neitsch et al. 2005. Wang and Melesse

(2005) evaluated the performance of snowmelt

hydrology of the SWAT model by simulating

streamflows for the Wild Rice River watershed

(located in the USA), and found that the SWAT

model had a good performance on simulating the

monthly, seasonal and annual mean discharges

and a satisfactory performance on predicting

the daily discharges. When analyzed alone, the

daily streamflows during the spring, which were

predominantly generated from melting snow,

could be predicted with an acceptable accuracy,

and the corresponding monthly and seasonal

mean discharges could be simulated very well.

The Gangotri glacier is located within the UGB.

Therefore, in order to include the contribution of

this glacier in the present model setup, discharge

data collected from a gauging site very close to

the snout of the glacier (Singh et al. 2006) was

added to the relevant subbasin.

7

FIGURE 3. The maps of the UGB used for modeling, with numbers and boundaries of subbasins used in hydrological simulations.

Note: FAO = Food and Agriculture Organization of the United Nations.

(a) Digital Elevation Model, SRTM

(c) Soil Map, FAO

(b) Land use map, Land Sat TM, 2003

(d) Mean Annual Precipitation (1971-2005)

8

Available Observed Time Series Data

The SWAT model requires time series of observed

climate data, including precipitation, minimum

and maximum temperature, sunshine duration,

wind speed and relative humidity. Table 1 lists the

climate stations used in the model. The locations

of stations are shown in Figure 1. The SWAT

model uses the data of a climate station nearest

to the centroid of each subcatchment as an input

for that subcatchment (Figure 3d).

Table 2 presents details of the flow stations

used for calibration and validation of the model.

Locations of the flow stations are shown in Figure

1. Due to restrictions on the distribution of Ganges

data from the Indian Central Water Commission

(CWC), only a very short time series of data at

some barrages were available. Although daily

observed flow data were available from Narora,

only monthly time series data were available for

the other sites. As the model works with daily

time steps, simulated daily flow values had to be

accumulated into monthly values for calibration

and validation. The existing dams, barrages and

irrigation deliveries were incorporated into the

model using available salient features from the

relevant barrage/dam authorities.

Station code1 Name Available record Available data type

R T S W H

42111 Dehradun2 1970-2005 x x x x x

42103 Ambala2 1970-2004 x x x x x

8207 Simla2 1989-2005 x x x x x

42140 Roorkee2 1970-1994;

2002-2005 x x x x x

42182 Delhi2 1970-2005 x x x x x

42366 Kanpur 1970-1974;

1986-1995 x x

42471 Fatehpur 1970-2005 x x

42189 Bareilly2 1970-2005 x x x x x

42260 Agra 1970-2005 x x

42262 Aligarh 1970-2005 x x

42143 Najibad2 1970-2005 x x x x x

42147 Mukteshwar2 1970-2005 x x x x x

42148 Pant Nagar2 1970-2005 x x x x x

42265 Mainpuri 1970-2005 x x x

42665 Shajapur 1970-2005 x

42266 Shahjahanpur 1970-2005 x x x x

Notes: 1 Station codes correspond to locations shown in Figure 1.

2 Station has large data gaps.

R = Precipitation; T = Minimum and maximum temperature; S = Sunshine duration; W = Wind speed; H = Relative humidity.

TABLE 1. Details of the meteorological data used.

9

��

Station code Location Catchment Available Type of data

area (km2) record

�� !�� "#$%$� �&� ��$$��'��$$*� +�� !

Flow_2 Narora 29,840 January 2000 - June 2005 Daily spill release from the dam

Flow_3 Kanpur 87,790 June 2003 - December 2005 Monthly spill release from the dam

� � � � /��0� �!��

Model Calibration and Validation

The period from January 1, 1970 to December

31, 1971 was used to warm-up the model. The

available data between 2000 and 2005 from each

station were divided into two sets, each of them

including the same number of daily observations.

The first and second sets were used for model

calibration and validation, respectively. Model

parameters were calibrated simultaneously for all

three flow stations.

The model performance was determined by

calculating the coefficient of determination (R2)

and the Nash-Sutcliffe Efficiency (NSE) criterion.

R2 and NSE values for each simulation are

presented in Table 3. The model performance

over cal ibrat ion and val idation periods is

acceptable, according to the model performance

ratings proposed by Liu and De Smedt (2004).

In addition, comparisons were made between

the measured and simulated annual water flow

volumes. The differences in volume ranged from

22 to 25% and from 7 to 9% during calibration

and validation periods, respectively.

Figure 4 shows observed precipitation,

observed and simulated flows for inflow into the

Bhimgoda barrage, outflow from Narora barrage

and outflow from Kanpur barrage. Although

flows are regulated at these sites, observed and

simulated hydrographs match very well. This adds

confidence to the results that are presented in the

following sections.

Downscaling of Climate Model Data

Climate data from PRECIS were used as input

to the SWAT hydrological model in order to

assess future river flow scenarios. PRECIS

is an atmospheric and land surface model

developed at the Meteorological Office Hadley

Centre, UK, for generating high-resolution

climate change information for many regions of

the world (Jones et al. 2004). It has a spatial

resolution of 0.22° x 0.22°. Climate data from the

GCM, Hadley Centre Coupled Model, version 3

(HadCM3) (Jones et al. 2004), were downscaled

with PRECIS for the UGB under A2 and B2

Special Report on Emission Scenarios (SRES)

scenarios (IPCC 2000) by IITM. A2 and B2 are

two climate change SRES scenarios studied

by the Intergovernmental Panel on Climate

Change (IPCC). A2 corresponds to a story line

of high population growth with slower per capita

economic growth and technological change, and B2

corresponds to a story line of moderate population

growth and economic development with less rapid

and more diverse technological change. PRECIS

data used in that study were extracted from 15

grid cells corresponding to the location of the

15 meteorological stations previously described.

These time series data cover the periods 1961-

1990 and 2071-2100, and include four variables:

precipitation, temperature, wind speed and relative

humidity. Although these PRECIS data result from a

downscaled global climate model which accounts for

regional climate and topographic characteristics, they

still exhibit discrepancies with regard to observed

meteorological data. For instance, the mean

absolute relative difference in annual precipitation

depths between PRECIS and observed time series

over the baseline period 1970-1990 is about 55%.

Therefore, PRECIS data were adjusted in such a

way that, at each of the 15 station locations, the

main statistical properties of adjusted PRECIS

output (mean and standard deviation) match those

of the historical data. This statistical downscaling

approach is described in Bouwer et al. (2004).

10

TA

BLE

3. A

sum

mary

of th

e S

WA

T m

odel p

erf

orm

ance e

valu

ation.

��

��2

��

��

��

�� 3

��

�� 4�

��

��

��0��

��

��

��0

�

S

tatistic C

alib

ration

V

alid

ation

C

alib

ration

Valid

ation

Valu

e

Ratings

Valu

e

Ratings

Observ

ed

Sim

ula

ted

Diffe

rence

Observ

ed

Sim

ula

ted

Diffe

rence

Flo

w_1

April 1,

January

1,

R2

0.8

6

> 0

.85

0.8

4

(0.6

5 -

0.8

5)

1,1

52

1,4

37

24.7

%

1,0

17

1,0

90

-7.2

%

(Bhim

goda)

2002 -

2004 -

D

ecem

ber

D

ecem

ber

Excelle

nt

V

ery

good

31, 2003

31, 2005

N

SE

0.6

7

(0.6

5 -

0.8

5)

0.8

1

(0.6

5 -

0.8

5)

V

ery

good

V

ery

good

Flo

w_2

January

1,

Jan 1

, R

2

0.9

2

(> 0

.85)

0.8

8

(> 0

.85)

905

1,1

03

21.8

%

694

754

8.6

%

(Naro

ra)

2000 –

2003 -

D

ecem

ber

June

Excelle

nt

E

xcelle

nt

31, 2002

30, 2005

N

SE

0.9

0

(> 0

.85)

0.8

5

(0.6

5 -

0.8

5)

E

xcelle

nt

V

ery

good

Flo

w_3

June 1

, June 1

,

R2

0.8

6

(> 0

.85)

0.6

0

(0.6

0 -

0.6

4)

756

923

22.1

%

622

663

6.7

%

(Kanpur)

2003 -

2005 –

O

cto

ber

D

ecem

ber

Excelle

nt

G

ood

3

1,

20

03

3

1,

20

05

and

N

SE

0.8

2

(0.6

5 -

0.8

5)

0.6

6

(0.6

5 -

0.8

5)

June 1

,

V

ery

good

V

ery

good

2004 -

O

cto

ber

31,

2004

Sta

tion

co

de

an

d

location

Ca

libra

tio

n

period

Va

lida

tio

n

period

11

FIGURE 4. Observed and simulated flows at (a) Bhimgoda barrage, (b) Narora barrage, and (c) Kanpur barrage.

0

400

800

1,200

1,600

2,0000

2,000

4,000

6,000

8,000

10,000

Jan-03 May-03 Oct-03 Mar-04 Aug-04 Jan-05 Jun-05 Nov-05

Prec

ipit

atio

n (m

m)

Flo

w (m

3 /s)

(c)

Precipitation Simulated flow Observed flow

0

400

800

1,200

1,600

2,0000

2,000

4,000

6,000

8,000

10,000

Jan-00 Oct-00 Aug-01 Jun-02 Apr-03 Feb-04 Dec-04 Oct-05

Prec

ipit

atio

n (m

m)

Flo

w (m

3 /s)

(b)


R2 = 0.92NSE = 0.90

R2 = 0.86NSE = 0.82 NSE = 0.66

Calibration Validation

R2 = 0.88NSE = 0.85

R2 = 0.60


R2 = 0.86NSE = 0.67

R2 = 0.84NSE = 0.81

0

400

800

1,200

1,600

2,0000

2,000

4,000

6,000

8,000

10,000

Jan-02 Jul-02 Jan-03 Jul-03 Jan-04 Jul-04 Jan-05 Jul-05 Jan-06

Prec

ipit

atio

n (m

m)

Flo

w (m

3 /s)

(a)



(a)

(b)

(c)

12

The period 1961-2008 was used to calculate the

standard deviation, and average of PRECIS and

observed data. For each of the four downscaled

climate variables, specific adjustment rules were

defined and are detailed below.

Precipitation

It is assumed that, for each month of the year,

the proportion of dry days (daily precipitation

= 0 mm) should be identical in both data sets

(PRECIS and observation). In order to meet this

hypothesis, the distribution of non-rainy days in

the original PRECIS data set was modified before

it was statistically downscaled, as described in

Bouwer et al. (2004). The successive steps of the

calculations are described below.

i) The average proportion (P) of rainy days

(> 0 mm) for each month of the year was

determined for both PRECIS and observed

data sets for each meteorological station.

P was found to be significantly higher for

PRECIS data, i.e., PRECIS data include fewer

dry days than observations. This difference

is explained by the fact that observed data

are point-based while PRECIS data are grid-

based (each cell extends over an area of 0.22

x 0.22 square degrees - about 25 x 25 km2).

Thus, the probability of having a dry day over

this area is much lower than that of having a

dry day at a point location.

ii) For each month o f the year, a da i ly

precipi tat ion threshold H (mm/day) is

determined in such a way that P% of PRECIS

daily precipitation values are higher than H.

iii) For each month of the year and for both data

sets (all observed values and PRECIS values

higher than H), the means, MOBSERVATION and

MPRECIS, and standard deviations, STOBSERVATION

and STPRECIS, respectively, of rainy day depths

(> H) were calculated.

iv) For each month of the year, adjusted PRECIS

daily values ‘y’ were calculated from the

original PRECIS data ‘x’ as shown in Equation

(2) below.

Temperature

As the PRECIS data set does not include

maximum and minimum temperature but only

daily averages, data adjustment was based on

mean observed daily temperature calculated from

observed maximum and minimum values. For

each month of the year, average and standard

deviation of daily temperatures were calculated for

observed and PRECIS data using available daily

values for the period 1961-2008. Data adjustment

was performed similarly to precipitation data,

although the first step of precipitation adjustment

(threshold definition) was not necessary in the

case of temperature.

Furthermore, the SWAT model required

maximum and minimum temperature as an

input and the adjusted PRECIS temperatures

were in daily average (xAVG). Therefore, the

observed daily data were used to calculate

monthly averages and standard deviations of

maximum, minimum and average temperature

(MOBS, MAX, MOBS, MIN, MOBS, AVG, STOBS, MAX, STOBS,

MIN, and STOBS, AVG), and then the following

equations were used to calculate the maximum

daily PRECIS temperature (TMAX) (Equation (3))

and minimum daily PRECIS temperatures (TMIN)

(Equation (4)).

)2(If x > H, then y = ( x - MPRECIS ) + MOBSERVATION

If x < H, then y = 0

STOBSERVATIONSTPRECIS

13

)3(Maximum daily PRECIS temperature, T MAX = ( x AVG - MOBS,AVG) + MOBS,MAX

STOBS, MAX

STOBS, AVG

)4(Minimum daily PRECIS temperature, T MIN = ( x AVG - MOBS,AVG) + MOBS,MIN

STOBS, MIN

STOBS, AVG

Wind speed

In the data set of observations, wind speed values

are measured 2 meters above the ground surface.

In contrast, PRECIS data correspond to wind

speeds 10 meters above the ground surface. This

difference in elevations results in higher PRECIS

wind velocities, as wind is slowed down by

frictional resistance close to the surface. As this

difference is observed for each month of the year,

a unique reduction coefficient (0.38) was applied

to all PRECIS daily values. This coefficient was

calculated by dividing the mean observed daily

wind speed value averaged over the period 1961-

2008 by that obtained from PRECIS data. Further

adjustment steps consisted of applying Equation

(1) to PRECIS wind speed daily values, similarly

to other climate variables.

Relative humidity

The adjustment of relative humidity is similar to

that of temperature and includes the same steps.

A further adjustment, when required, consisted

of replacing values exceeding 100% by ‘100%’.

Such cases (less than 1% of adjusted values),

was due to the following reasons: in some cases,

the range of observed relative humidity values

over the baseline period was found to exceed

that of the PRECIS data. This resulted in higher

monthly averages and/or standard deviations for

observed values, in comparison with those of

PRECIS data. The latter, when used for the data

adjustment, could produce daily relative humidity

values exceeding 100%.

Scenario Setting

Four scenarios were simulated. In scenario 1,

the water infrastructure system up until the year

2005 was included. This scenario includes water

abstractions from dams and barrages. The Tehri

Dam (Figure 1), which became operational in

2008, is not included in this scenario. Irrigated

crops such as rice, wheat, corn, finger millet,

sugarcane and potato represent the major crop

types during present conditions. Scenario 2

represents the ‘naturalized’ condition without any

artificial flow regulation. Simulated flows for this

scenario were produced after having removed

all water infrastructures from the calibrated

model. In this scenario, farming areas were

characterized by rainfed crops such as mung

bean and wheat. Most of the non-agricultural

land is covered by natural forest. Scenarios 3

and 4 correspond to the present conditions of

water infrastructure development, abstraction

(up until 2005) with CC scenario adjusted from

PRECIS time series data under SRES A2 and

B2, respectively. Table 4 provides a detailed

description of the scenarios.

TABLE 4. Description of simulated scenarios.

No. Water infrastructure development Climate input data

1 Present (as of 2005) Observed data 1971-2005

2 Naturalized conditions Observed data 1971-2005

3 Present (as of 2005) Adjusted PRECIS data over period 2071-2100 under A2 scenario

4 Present (as of 2005) Adjusted PRECIS data over period 2071-2100 under B2 scenario

14

Results and Discussion

Water Balance Under Present Conditions

and Natural Conditions (Scenarios 1 and 2)

Figure 5 shows the annual average water balance

for all 21 subbasins of the UGB, and Figure

6 shows the mean monthly water balance of

the whole basin, both under Scenario 1. Four

hydrological components are considered, i.e.,

precipitation, actual evapotranspiration (ET), net

water yield (which is a routed runoff from the

subbasin) and balance closure. The term ‘balance

closure’ includes groundwater recharge, change

in soil moisture storage in the vadose zone and

model inaccuracies.

Annual average precipitation, actual ET and

net water yield of the whole basin were 1,192,

416 and 615 mm, respectively. However, there

was a large variation in the spatial distribution of

these components. Precipitation, ET and water

yield were found to be higher in the forested and

mountainous upper areas. In the upper subbasins,

water yield is higher than ET. However, in some

of the lower subbasins dominated by agriculture,

ET values were higher than water yield. Water

balances from the lower part of the catchment,

containing irrigated areas, are affected by water

regulation through barrages, dams and canals.

The large network of canals is transferring water

from one subbasin to irrigate crops in another

subbasin. Therefore, it is difficult to determine

the accuracy of the runoff calculations from each

subbasin. However, the ET and precipitation

figures are useful in characterizing the water

availability and use in each of the subbasins. The

maximum precipitation of 2,504 mm occurred in

subbasin 3 and minimum precipitation of 536 mm

occurred in subbasins 8 and 11 (see also Figure

3(d)). Furthermore, the maximum ET of 671 mm

occurred in subbasin 7 and the minimum ET of

177 mm occurred in subbasin 10.

The mean monthly results from 1971 to 2005

(Figure 6) show that there are large temporal

variations in the water balance components. The

maximum precipitation of 338 mm occurred during

August and a minimum of 7 mm occurred in

November. Similarly, water yields are also much

higher during the monsoon months as compared to

the dry season. ET, however, which is more related to

land cover, was found to be lowest during the winter

months, i.e., November-January (post-rice harvest).

-3,000

-1,500

0

1,500

3,000

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21

An

nu

al a

vera

ge

wat

er b

alan

ce(m

m)

Subbasin number

Precipitation Actual ET Net water yield Balance closure

FIGURE 5. Annual average water balance results of model simulation at the subbasin level (1971-2005). The subbasin

numbers are given in Figure 3.

15

Simulation of Natural Flow Conditions for

the Four EF Sites (Scenario 2)

Simulating non-regulated, pre-development

flow regimes is important to determine the

re ference hydro logy, against which f low

changes in a basin can be measured. Strictly

speaking, this constitutes the assessment

of water actually available for all uses and

development – a starting point in sustainable

basin planning and management. Natural flow

simulations are also required to assess EF

– the flow regimes that are required for the

ecological health of the river. Results from

the EF assessment for the UGB are beyond

the scope of this paper and are discussed in

another report (WWF 2011).

Natural flows for four locations (sites EF1-

EF4, coordinates in Table 5) are presented in the

following sections. Although natural flows have

been simulated for the whole basin, these four

locations were chosen for this study as they are

representative of different agroecological zones

in the river stretch used for this study. These

locations were also sites for the EF assessment

study and are, therefore, referred to as EF sites

in this report.

Simulated daily flow data at the four EF sites

under scenarios 1 and 2 were summed up to

monthly and annual time steps, and are presented

in the tables and figures below. As already

mentioned above, the modeling period only went

up to 2005, so the effect of the Tehri Dam, which

became operational in 2008, was not considered.

Therefore, for site EF1, only natural flows have

been reported because water is neither stored in

dams nor abstracted for agriculture upstream of

this site.

-400

-200

0

200

400

Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec

Mea

n m

on

thly

wat

er b

alan

ce(m

m)

Month

Precipitation Actual ET Net water yield Balance closure

FIGURE 6. Mean monthly water balance results of model simulation (1971-2005) for the entire UGB - at Kanpur.

TABLE 5. Location and names of representative sites along the UGB (see also Figure 1).

Site code Name Latitude Longitude

EF1 Kaudiyala Rishikesh 30°04’29” N 78°30’09” E

EF2 Narora 29°22'22” N 78°2'20” E

EF3 Kachla Bridge 27°55’59” N 78°51’42” E

EF4 Bithur (Kanpur) 26°36'59” N 80°16'29” E

16

Figure 7 shows the plots of annual water

flow volume and monthly water flow volume at

site EF4, Bithur (Kanpur), which is located near

the outlet of the UGB. Table 6 shows simulation

flow results for the four EF sites including

simulated present f lows for EF sites 2-4.

Comparison between natural and present flows

showed that, on average, the present annual

water volume is 7%, 2% and 8% lower than

in the natural conditions at EF sites 2, 3 and

4 (Narora, Kachla Bridge and Bithur (Kanpur)),

respectively. At all sites, the percentage of flow

reduction is highest during the dry months as

water is being withdrawn for irrigation. At Narora

(site EF2), maximum flow reduction is 70% in

February. Similarly, at site EF3, the maximum

flow reduction is 35% in February and at site

EF4, the maximum flow reduction is 58% also

in February (Table 6). In April, and some other

months, mean flow volume of the present

conditions exceeds the flow volume of the natural

conditions. This is caused by the basin flow

transfer from Ramganga Dam. Although high

percentages of flow reduction occur in the dry

season, the contribution of flow in this season to

annual flow is low (less than 10%).

FIGURE 7. (a) Annual flow totals, and (b) average monthly flow distribution for Bithur (Kanpur).

0

2,000

4,000

6,000

8,000

10,000

12,000

14,000

16,000

18,000

Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec

Vo

lum

e (M

CM

)

Month

EF4 - Bithur (Kanpur)

Natural monthly flow volume Present monthly flow volume

(a)

0

20,000

40,000

60,000

80,000

100,000

1971

1973

1975

1977

1979

1981

1983

1985

1987

1989

1991

1993

1995

1997

1999

2001

2003

2005

Vo

lum

e (M

CM

)

Year


Annual natural flow volume Annual present flow volume

(b)

(a)

(b)

17

EF

sites

Annual

sim

ula

ted f

low

A

nnual

Perc

enta

ge o

f m

onth

ly r

eduction i

n f

low

s (

%)

v

olu

me (

MC

M)

re

duction

in f

low

s

(volu

me

N

atu

ral

P

resent

(%))

Jan

Feb

Mar

Apr

May

Jun

Jul

Aug

Sep

Oct

Nov

Dec

EF

1

(Kaudiy

ala

/

Ris

hik

esh)

38,4

45

38,4

45*

0 (

0%

) 0

0

0

0

0

0

0

0

0

0

0

0

EF

2

(Naro

ra)

42,6

08

39,6

32

2,9

76 (

7%

) 68

70

44

-10

30

24

0

0

2

4

19

18

EF

3

(Kachla

Bridge)

42,9

09

42,1

65

774 (

2%

) 27

35

14

-60

7

16

-1

-1

0

2

12

18

EF

4

(Bithur

(Kanpur)

) 57,0

61

52,2

68

4,7

93 (

8%

) 44

58

35

-44

12

31

11

7

5

2

4

10

TA

BLE

6. S

imula

ted flo

w r

esults a

t E

F s

ites.

Not

e: *

No r

educt

ion c

om

pare

d w

ith n

atu

ral c

onditi

ons,

beca

use

there

was

no w

ate

r use

upst

ream

of th

is s

ite d

uring the s

tudy

period.

18

Figure 8 shows the comparison of water

yield (runoff from subbasin) distribution at the

subbasin level for the natural condition as well

as the present condition. As can be expected,

water yield or runoff, from the upper forested

subbasins, have not changed. However, there

are reduced flows during the present condition

from the lower subbasins, mainly due to water

withdrawals for agricultural production. There are

a few subbasins in the lower catchment where

water yields have increased and this is a result

of water transfers from the Upper and Madhya

Ganga canals.

Figure 9 shows flow duration curves (FDC)

in present and natural conditions at the EF sites.

These curves indicate that flows are lower in

the present condition in comparison with natural

conditions.

FIGURE 8. Percentage change in mean annual net water yields at present condition in comparison to natural condition of model

simulation at subbasin level (i.e., present-natural).

19

FIGURE 9. Flow duration curves (FDC) for (a) Kaudiyala/Rishikesh, (b) Narora, (c) Kachla Bridge, and (d) Bithur (Kanpur) sites.

100

1,000

10,000

0 20 40 60 80 100

Flo

w (m

3 /s)

Flo

w (m

3 /s)

Percentage of exceedence (%)

EF1 - Kaudiyala/Rishikesh

10

100

1,000

10,000

0 20 40 60 80 100


EF2 - Narora

Natural Present

100

1,000

10,000

0 20 40 60 80 100

Flo

w (m

3 /s)

Flo

w (m

3 /s)


EF3 - Kachla Bridge

Natural Present

10

100

1,000

10,000

0 20 40 60 80 100



Natural Present

(a) (b)

(c) (d)

Table 7 shows the difference in flow rates

between simulations of natural and present

conditions for 40% to 90% of percentile of

exceedence. Usually, water-related infrastructure

such as hydropower plants or irrigation systems

are designed taking into consideration a design

discharge corresponding to 40% to 90%. The

difference in flows affects any future planning

of water resources development as well as

allocations for environmental flows in the river.

TABLE 7. Difference in flows between simulations of the natural and present conditions at EF sites.

�� 3��0��'&��4�3�3/s)

Percentage of exceedence EF2 (Narora) EF3 (Kachla Bridge) EF4 (Bithur (Kanpur))

40% 232 99 156

50% 120 28 115

60% 93 17 48

70% 89 25 54

80% 114 41 68

90% 126 37 88

20

��

Characteristics Under Different Simulation

Scenarios Including CC (Scenarios 1-4)

Five indicators of hydrological variability were

derived from simulated time series: mean daily dry

season flows, mean daily wet season flows, the

date of maximum flow, the date of minimum flow

and the number of reversals reflecting the rate of

change in river discharge.

Hydrological Variability and Alteration Caused by Dams Under Present Climate Conditions

The analysis first focuses on flows simulated from

observed precipitation, either under the present

conditions of water infrastructure development

or in naturalized conditions. Figure 10 displays

mean dry and wet season discharges computed

from mean monthly discharge values. In natural

conditions, dry season flows remain stable all

along the river stream, with values ranging from

302 to 340 m3/s. This flow regime indicates that

river flow mostly originates from the melting of

glacier and snow cover while the flow contribution

from the water table drainage is negligible. During

the wet season, mean daily discharge consistently

increases from upstream to downstream, reflecting

the successive flow contributions of the tributaries,

and collecting surface runoff produced by

monsoon rainstorms. While the impact of water

infrastructure remains moderate during the wet

season (flows under present condition are 6%

lower than flows produced in natural conditions),

relative flow changes during the dry season are

of a higher magnitude, especially at EF sites 2,

3 and 4, as no dam exists in the EF1 headwater

catchment. Between the sites EF1 and EF4, the

dams have induced a 25% flow decrease as a

result of dry season irrigation.

Figure 11 displays the mean Julian date of

1-day minimum and maximum flow at four EF

sites, under natural and present conditions. In

natural conditions, 1-day minimum and maximum

flow conditions are delayed downstream. This

shift in the flow regime results from the difference

in the onset of the monsoon between upstream

and downstream parts of the basin. First rains

of the wet season occur in the upper part

of the basin (Figure 12). Under the present

conditions, the downstream shift in the date

of the 1-day maximum flow is similar to that

observed under natural conditions. In contrast,

the 1-day minimum flow does not follow this

pattern. Dates at sites EF1 and EF3 are similar

(dates = 22) while the 1-day minimum flow

occurs much earlier at sites EF2 and EF4 (date

= 1 for EF2; date = 2 for EF4). This alteration of

the natural flow regime results from the operation

of the dams located upstream of the sites EF2

and EF4, storing flows at the beginning of the

year. The impact of these dams on the date of

the 1-day maximum flow is imperceptible, as the

range of flow variations caused by the operation

of the dams is much lower than the mean river

discharge during this period of the year.

FIGURE 10. Mean dry and wet season daily discharge for the period 1971-2005 at four EF sites, under naturalized (natural)

and present conditions.

0

50

100

150

200

250

300

350

400

EF1 EF2 EF3 EF4

Mea

n d

isch

arg

e (m

3 /s)

Mea

n d

isch

arg

e (m

3 /s)

Dry season

0

500

1,000

1,500

2,000

2,500

3,000

3,500

4,000

EF1 EF2 EF3 EF4

Wet season

Natural

Present

21

FIGURE 11. Timing of annual extreme water conditions.

0

20

40

60

80

100

120

EF1 EF2 EF3 EF4

Julia

n d

ate

Date of minimum flow

220

222

224

226

228

230

232

234

236

238

240

EF1 EF2 EF3 EF4

Julia

n d

ate

Date of maximum flow

Natural

Present

The number of reversals is calculated by

dividing hydrological records into ‘rising’ and

‘falling’ periods in which daily changes in flows

are either positive or negative, respectively.

The annual average number of reversals

indicates whether a flow regime is influenced

only by precipitation input or includes anthropic

alterations. Figure 13 displays the mean annual

number of reversals at four EF sites, under

natural and present condit ions. In natural

conditions, the number of reversals decreases

downstream, reflecting the lower temporal

variability of the natural flow regime, mostly

FIGURE 12. Mean observed precipitation (1971-2005) in the subcatchments of four EF sites.

0

50

100

150

200

250

1 3 5 7 9 11 13 15 17 19 21 23 25 27 29 31 33 35

Prec

ipit

atio

n d

epth

s (m

m)

10-day periods

EF1 EF2

EF3 EF4

caused by the mainstream’s integration of flow

fluctuations of the tributaries. Under present

conditions, the number of reversals at site

EF2 is greater than the number of reversals

recorded at site EF1. The increase in the flow

variability between sites EF1 and EF2 reflects

the great impact of dams located between the

two stations. As flows are still moderate in the

upper part of the basin, the operation of dams

can significantly alter the natural river regime,

unlike in the downstream areas where the

greater river discharge is less impacted by dam

operation.

22

FIGURE 13. Mean annual number of reversals recorded at four EF sites and at the basin outlet, under natural and present conditions.

0

20

40

60

80

100

120

EF1 EF2 EF3 EF4

Nu

mb

er o

f rev

ersa

ls/y

ear

Number of reversals

Natural

Present

Hydrological Changes Caused by ‘Climate Change’ Scenarios Under Present Water Infrastructure Development

Figure 14 displays the mean dry and wet season

flows at the four EF sites under present climate

conditions, and climate change projections from

the PRECIS RCM under A2 (PRECIS-A2) and

B2 (PRECIS-B2) scenarios. Upstream of site

EF3, dry and wet season flows produced by

PRECIS under the A2 scenario, precipitation is

lower than that in present climate conditions.

This tendency inverts downstream of the site

EF3, with higher dry and wet season flows at

site EF4 under PRECIS-A2 precipitation, in

comparison with flows produced by present

climate conditions. Similar patterns are observed

for precipitation (Figure 15), suggesting that

spatial variations in flow patterns originates

from the precipitation distribution over the

subcatchments of the four EF sites. Flows

produced by PRECIS-B2 precipitation are

systematically higher and lower than those

produced by PRECIS-A2 precipitation during

the dry season and wet season, respectively.

These flow differences most likely result from

the difference of PRECIS-A2 and PRECIS-B2

precipitation as displayed in Figure 15.

FIGURE 14. Mean dry and wet season daily discharge under present climate conditions (for the period 1971-2005), and for future

climate conditions under scenarios A2 and B2 (for the period 2071-2100).

0

50

100

150

200

250

300

350

400

450

EF1 EF2 EF3 EF4

Mea

n d

isch

arg

e (m

3 /s)

Dry season

0

500

1,000

1,500

2,000

2,500

3,000

3,500

4,000

EF1 EF2 EF3 EF4

Mea

n d

isch

arg

e (m

3 /s)

Wet season

Presentclimate

PRECIS-A2

PRECIS-B2

23

FIGURE 15. Mean dry and wet season precipitation under present climate conditions (observed precipitation averaged over period

1971-2005), and for future climate conditions under scenarios A2 and B2 (for the period 2071-2100).

0

100

200

300

400

500

600

700

EF1 EF2 EF3 EF4

Dry season

Present climate

PRECIS-A2

PRECIS-B2

0

200

400

600

800

1,000

1,200

1,400

1,600

1,800

EF1 EF2 EF3 EF4

Mea

n a

nn

ual

pre

cip

itat

ion

(mm

)

Mea

n a

nn

ual

pre

cip

itat

ion

(mm

)

Wet season

Figure 16 displays the average Julian dates

of minimum and maximum 1-day flow at four EF

site under present and future climate conditions.

The dates of minimum 1-day flow are highly

variable among stations and between different

climate scenarios. This variability is caused by the

operation of the dams, as the volume of controlled

outflow is of the same order of magnitude as

natural flow. As a result, a slight change in the

dam outflow results in a significant change in the

date of the minimum flow. The same phenomenon

explains the delay of the minimum flows under

scenarios A2 and B2. Julian dates of maximum

flows are delayed downstream as a result of

the delay in the onset of the monsoon in the

lower parts of the basin (Figure 12 and Figure

17). Maximum flows simulated from PRECIS-A2

precipitation occur later than maximum flows

simulated from PRECIS-B2 precipitation. This time

lag is caused by the slight difference in the timing

of the wettest period, which occurs earlier in the

case of PRECIS-B2.

FIGURE 16. Occurrence of annual extreme flow conditions.

0

20

40

60

80

100

120

140

EF1 EF2 EF3 EF4

Julia

n D

ate

Julia

n D

ate

Date of minimum flow

215

220

225

230

235

240

EF1 EF2 EF3 EF4

Date of maximum flow

Present

climate

PRECIS-A2

PRECIS-B2

24

Figure 18 displays the number of reversals

at four EF sites. At sites EF1, EF2 and EF3,

the number of reversals under climate change

conditions (PRECIS-A2 and PRECIS-B2) is lower

than the number of reversals observed under

present climate conditions. At site EF4, this

tendency reverses and the number of reversals

becomes greater under climate change conditions.

The greater number of reversals under climate

change conditions in the downstream part of

the catchment is caused by the higher variability

of precipitation. At each EF site, the number

of reversals is greater under PRECIS-B2 in

comparison with PRECIS-A2. This difference

results from the greater temporal variability of

PRECIS-B2 precipitation.

FIGURE 18. Mean annual number of reversals recorded at four EF sites, under present and future climate conditions.

FIGURE 17. Mean precipitation in the subcatchments of four EF sites, as predicted by PRECIS A2 and B2 scenarios. Averages

computed over the period 2071-2100.

0

20

40

60

80

100

120

140

160

180

200

1 3 5 7 9 11 13 15 17 19 21 23 25 27 29 31 33 35

10-d

ay p

reci

pit

atio

n d

epth

(mm

)

10-d

ay p

reci

pit

atio

n d

epth

(mm

)

10-day period 10-day period

PRECIS-A2EF1

EF2

EF3

EF4

0

20

40

60

80

100

120

140

160

180

200

1 3 5 7 9 11 13 15 17 19 21 23 25 27 29 31 33 35

PRECIS-B2EF1

EF2

EF3

EF4

0

20

40

60

80

100

120

EF1 EF2 EF3 EF4

Julia

n d

ate

Number of reversals

Present climate

PRECIS-A2

PRECIS-B2

25

Conclusions

This report explains the results of the first (known

to authors) attempt to analyze the impacts of

water infrastructure development in the entire

UGB, by comparing flow changes under natural

and present conditions. The analysis shows that,

on average, annual flows at present are 2-8%

lower than under naturalized conditions. Higher

flow reduction in the dry season (up to 70% in

February) is detected, compared to just a small

percentage change in the wet season. Therefore,

various dams and barrages constructed to date

have reduced mainly the flows during the dry

season – when irrigation water demands are

the highest. Flow regulation through dams and

barrages has also changed the timing of annual

extreme water conditions such as the date of

minimum and maximum flows. The change in

the timing of the minimum flow date is, however,

affected more than the maximum flows. Future

water resources development plans need to take

this into serious consideration in order to avoid

further detrimental impacts on the river ecology.

Also, the study simulated the impacts of

CC on water infrastructure development in

the UGB. The results suggest that both dry

and wet season flows under CC scenario A2

(scenario corresponding to high population

growth with slower per capita economic growth

and technological change) are lower than that

under present climate conditions at upstream

locations, but higher at downstream locations

and at the basin outlet. Flows simulated under

CC scenario B2 (corresponding to moderate

population growth and economic development

with less rapid and more diverse technological

change) are found to be higher during the dry

season, and lower during the wet season than

that under CC scenario A2. Under CC scenario

B2, the timing of the maximum flow period is

earlier than that under present conditions. This

basically means that the monsoon might start

earlier. Furthermore, greater temporal variability of

precipitation was found in the lower basin under

both the A2 and B2 scenarios. All these results

are very relevant to future water management

in the basin. In the upper parts of the basin,

especially in the Uttarkhand District, plans for

further hydropower development are underway.

Decrease in precipitation and flows will affect

water availability in the planned projects. Similarly,

change in the timing of the monsoon as well as

increased variability in precipitation, especially in

the lower parts of the basin, will affect the current

irrigation water regulation practices. Therefore,

some adjustment to agricultural practices such as

early sowing might be necessary if the projected

changes under the B2 scenario become a reality.

The present modeling study did not consider

scenarios of future water resources development

under future climate scenarios. As mentioned

above, in the UGB, especially in upper parts of

the basin, several hydropower dams are being

planned or operationalized. The impacts of the

already constructed Tehri Dam are now coming

into effect. The combined impact of future water

infrastructure development and climate on river

flow in the UGB and the availability of water for

agriculture and other uses, as well as the impacts

of CC on operation of infrastructure itself, is a

subject of a subsequent ongoing study.

It could be argued that precipitation scenarios

used to anticipate hydrological change under CC

are not reliable as they originate only from one

climate model: PRECIS RCM forced by HadCM3.

It is now accepted that climate models are not

able to accurately simulate precipitation, mostly

because of their inability to simulate actual climate

dynamics. For instance, Kingston et al. (2011)

showed that uncertainty in precipitation is the

main source of error in hydrological projections.

Even the use of averaged precipitations projected

by several climate models cannot reduce this

uncertainty as the variability between different

climate projections from different models is

high. However, Rupa Kumar et al. (2006), who

assessed the biases in PRECIS simulation over

India by comparing simulated and observed

precipitation, found that this RCM is able to

reasonably predict the climate over India, both in

terms of means and extremes. A second source of

26

uncertainty in the PRECIS precipitation projections

originates from the SRES scenarios, associated

with specific gas emission conditions that may

result in various precipitation conditions. In the

present study, two very contrasting scenarios

are used to cover the whole range of possible

precipitation changes. Consequently, even if the

CC projections used in this study are biased

because of the use of only one climate model, it

is moderated by the use of the two contrasting

SRES emission scenarios - A2 and B2.

The main constraint in this study (as well as

in all research carried out on water resources in

the Ganga Basin) is the availability of observed

data (on climate, hydrology, etc.). This limitation

severely affects model calibration and validation,

and, in the end - simulations of future scenarios.

The authori t ies responsible for observed

hydrological data management and sharing should

seriously consider opening their archives for water

research needs. Without more open policies on

observed data access, proper planning of water

resources development in the Himalayan parts

of India is impossible. The uncertainty in climate

projections is another major issue associated

with studies like this one. Improvement in CC

projections as well as access to more climate

data would certainly enhance the accuracy of the

simulations, and in the end – planning for the

future.

27

References

Ahmad, Q.K.; Ahmed, A.U.; Khan, H.R.; Rasheed, K.B.S. 2001. GBM regional water vision: Bangladesh

perspectives. In: Ganges-Brahmaputra-Meghna region: A framework for sustainable development, (eds.), Ahmad,

Q.K.; Biswas, A.K.; Rangachari, R.; Sainju, M.M. Dhaka: University Press Limited. pp. 31-78.

Akhtar, M.; Ahman, N.; Booij, M.J. 2009. Use of regional climate model simulations as input for hydrological models

for the Hindukush-Karakorum-Himalaya region. Hydrology and Earth System Sciences 13: 1075-1089.

Arnell, N.W. 2003. Relative effects of multi-decadal climatic variability and changes in the mean and variability of

�� 0��!�� !8��0�0�� Journal of Hydrology 270(3-4): 195–213.

Arnold, J.G.; Srinivasan, R.; Muttiah, R.S.; Williams, J.R. 1998. Large area hydrologic modeling and assessment.

Part I: Model development. Journal of the American Water Resources Association 34(1): 73–89.

Benestad, R.E.; Hanssen-Bauer, I.; Chen, D. 2008. Empirical-statistical downscaling�� 9 �!�&��8� :�� 9� �� ?��

Publishing Co. pp. 300.

Bergström, S.; Carlsson, B.; Gardelin, M.; Lindström, G.; Pettersson, A.; Rummukainen, M. 2001. Climate change

impacts on runoff in Sweden-assessments by global climate models, dynamical downscaling and hydrological

modelling. Climate Research 16: 101-112.

�� #��@�B��C��#�2��$��D��! �� &�� '0��!�� F�� G ��#�

Bangladesh. Water International 36(3): 357-369.

�� H#� +�B�� $$*�� J�&�� !� �� !� �� &�� 0� ��

Journal of Hydrology 303(1-4): 176-198.

Bouwer, L.M.; Aerts, J.C.J.H.; van de Coterlet, G.M.; van de Giesen, N.; Gieske, A.; Mannaerts, C. 2004. Evaluating

downscaling methods for preparing Global Circulation Model (GCM) Data for Hydrological Impact Modelling. In:

Climate Change in Contrasting River Basins, (eds.) Aerts, J.C.J.H.; Droogers, P. Wallingford, UK: CABI Publishing.

de Condappa, D. 2009. WEAP modeling for Indus and Ganges basin. Unpublished Report.

��#�D�B�@�K ��#�L�F�� $$N��O� �!� �! �� 0�� 0�0��

northwest England. Climatic Change 80(3-4): 337-367.

Garg, K.K.; Bharati, L.; Gaur, A.; George, B.; Acharya, S.; Jella, K.; Narasimhan, B. 2011. Spatial mapping of

agricultural water productivity using the SWAT model in Upper Bhima Catchment, India. Irrigation and Drainage

3�� ?��&0�� 4�

Gosain, A.K.; Rao, S.; Basuray, D. 2006. Climate change impact assessment on hydrology of Indian river basins.

Current Science 90(3): 346-353.

Guo, S.; Wang, J.; Xiong, L.; Yin, A.; Li, D. 2002. A macro-scale and semi-distributed monthly water balance model

to predict climate change impacts in China. Journal of Hydrology 268(1-4): 1-15.

Gutowski, W.J.; Decker, S.G.; Donavon, R.A.; Pan, Z.; Arritt, R.W.; Takle, E.S. 2003. Temporal-spatial scales of

observed and simulated precipitation in central U.S. climate. Journal of Climate 16(22): 3841-3847.

Hay, L.E.; Clark, M.P. 2003. Use of statistically and dynamically downscaled atmospheric model output for hydrologic

simulations in three mountainous basins in the western United States. Journal of Hydrology 282(1-4): 56-75.

Hay, L.E.; Clark, M.P.; Wilby, R.L.; Gutowski, W.J.; Leavesley, G.H.; Pan, Z.; Arritt, R.W.; Takle, E.S. 2002. Use of

regional climate model output for hydrological simulations. Journal of Hydrometeorology 3(5): 571–590.

Jian, J.; Webster, P.J.; Hoyos, C.D. 2009. Large-scale controls on Ganges and Brahmaputra river discharge on

intraseasonal and seasonal time-scales. Quarterly Journal of the Royal Meteorological Society 135(639): 353-370.

IPCC (Intergovernmental Panel on Climate Change). 1996. Climate change 1995: Impacts, adaptation, and mitigation �� report of the Intergovernmental Panel on Climate Change, (eds.), Watson, R.T.; Zinyowera, M.C.; Moss, R.H.

Cambridge, UK: Cambridge University Press.

28

IPCC. 2000. IPCC special report on emission scenarios. Summary for policymakers. A special report of working

group III of the Intergovernmental Panel on Climate Change (IPCC. Geneva, Switzerland: Intergovernmental

Panel on Climate Change (IPCC). 20p.

Jha, M.K. 2011. Evaluating hydrologic response of an agricultural watershed for watershed analysis. Water 3(2):

604-617.

Jones, R.G.; Noguer, M.; Hassell, D.C.; Hudson, D.; Wilson, S.S.; Jenkins, G.J.; Mitchell, J.F.B. 2004. Generating high resolution climate change scenarios using PRECIS��+��P�?��D��L��#��/��#�OK��Q$&��

��&8RR&�� ?��R��R2G�LJ9�D��C�&��3��P��Q#��$��4�

Kingston, D.G.; Thompson, J.R.; Kite, G. 2011. Uncertainty in climate change projections of discharge for the

Mekong River Basin. Hydrology of Earth System Sciences 15(5): 1459-1471.

Leander, R.; Buishand, T.A. 2007. Resampling of regional climate model output for the simulation of extreme river

��Journal of Hydrology 332: 487-496.

Liu, Y.B.; De Smedt, F. 2004. WetSpa Extension, �� Department of Hydrology and

Hydraulic Engineering, Vrije Universiteit Brussel, Belgium. 108p.

Mirza, M.M.Q. (ed.). 2004. The Ganges water diversion: Environmental effects and Implications. Dordrecht, the

Netherlands: Kluwer Academic Publishers.

Neitsch, S.L.; Arnold, J.G.; Kiniry, J.R.; Williams, J.R. 2005. Soil and Water Assessment Tool – Theoretical documentation (version 2005). Temple, Texas: Grassland, Soil and Water Research Laboratory, Agricultural

Research Service, Blackland Research Center, Texas Agricultural Experiment Station.

Pilling, C.G.; Jones, J.A.A. 2002. The impact of future climate change on seasonal discharge, hydrological processes

��/�� O&&��:��/&� ��#�� ':��Hydrological Processes 16(6): 1201-1213.

G��#�+�+��$$U�� J��!��F��!�� !��8�L�� &� �� ! ��&��Water Policy 11(2): 168-190.

Rupa Kumar, K.; Sahai, A.K.; Krishna Kumar, K.; Patwardhan, S.K.; Mishra, P.K.; Revadekar, J.V.; Kamala, K.;

Pant, G.B. 2006. High-resolution climate change scenarios for India for the 21st century. Current Science 90(3):

334-345.

Seidel, K.; Martinec, J.; Baumgartner, M.F. 2000. Modelling runoff and impact of climate change in large Himalayan basins. International Conference on Integrated Water Resources Management (ICIWRM), December 19-21,

2000, Roorkee, India.

9 �!�#� 2�@� D�� #� O�K�@� K0��#� V�� $$%�� +�� !� �� &��

Gangotri Glacier basin, Himalayas. Journal of Geophysical Research 53(2): 309-322.

Singh, P.; Haritashya, U.K.; Kumar, N.; Singh, Y. 2006. Hydrological characteristics of the Gangotri Glacier, central

Himalayas, India. Journal of Hydrology 327(1-2): 55-67.

Srinivasan, R.; Ramanarayanan, T.S.; Arnold, J.G.; Bednarz, S.T. 1998. Large area hydrologic modeling and

assessment Part II: Model application. Journal of the American Water Resources Association 34(1): 91-101.

Verghese, B.G.; Iyer, R.R. 1993. Harnessing the eastern Himalayan rivers: Regional cooperation in South Asia.

New Delhi: Konark Publishers.

Wang, X.; Melesse, A.M. 2005. Evaluation of the SWAT Model’s snowmelt hydrology in a Northwestern Minnesota

watershed. Transactions of the American Society of Agricultural Engineers 48(4): 1359-1376.

Wilby, R.L.; Hay, L.E.; Leavesley, G.H. 1999. A comparison of downscaled and raw GCM output: Implications for

climate change scenarios in the San Juan River Basin, Colorado. Journal of Hydrology 225(1-2): 67-91.

::�� 3:�� : �� 0�� V��0�4'J�� $�� O&&�� F��!��

New Delhi, India: World Wide Fund for Nature (WWF)-India. 21p. Available at ��!!�� !��"��!�� "��!��#�� " �� $! (accessed on October 31, 2011).

Electronic copies of IWMI's publications are available for free.

Visit

www.iwmi.org/publications/index.aspx

IWMI Research Reports

142 The Impacts of Water Infrastructure and Climate Change on the Hydrology of the Upper Ganges River Basin. Luna Bharati, Guillaume Lacombe, Pabitra

Gurung, Priyantha Jayakody, Chu Thai Hoanh and Vladimir Smakhtin. 2011.

141 Low-Cost Options for Reducing Consumer Health Risks from Farm to Fork Where Crops are Irrigated with Polluted Water in West Africa. Philip Amoah,

�� !��"��#�� $��%�

Konradsen. 2011.

140 An Assessment of Crop Water Productivity in the Indus and Ganges River Basins: Current Status and Scope for Improvement��&��%�#��Sharma, Mir Abdul Matin, Devesh Sharma and Sarath Gunasinghe. 2010.

139 Shallow Groundwater in the Atankwidi Catchment of the White Volta Basin: Current Status and Future Sustainability. Boubacar Barry, BenonyKortatsi,

'��$��'��%��(��)��*��

Joost van den Berg and Wolfram Laube. 2010.

138 Bailout with White Revolution or Sink Deeper? Groundwater Depletion and Impacts in the Moga District of Punjab, India. Upali A. Amarasinghe, Vladimir

+��+��(��,��%��-./.�

137 Wetlands, Agriculture and Poverty Reduction. Matthew McCartney, Lisa-Maria

��+��+��+��+��0� ��+�� -./.�

136 Climate Change, Water and Agriculture in the Greater Mekong Subregion��1��'��)��#��2��3�� (��!��!� ��

Vladimir Smakhtin, Diana Suhardiman, Kam Suan Pheng and Choo Poh Sze.

2010.

135 Impacts of Climate Change on Water Resources and Agriculture in Sri Lanka: A Review and Preliminary Vulnerability Mapping��(��,��%��4��+��)��#��$��-./.�

134 Evaluation of Current and Future Water Resources Development in the Lake Tana Basin, Ethiopia. Matthew McCartney, Tadesse Alemayehu, Abeyu Shiferaw

and Seleshi Bekele Awulachew. 2010.

133 Mapping Drought Patterns and Impacts: A Global Perspective�� (��,��%��4��+��(��'��%��-..5�

Postal AddressP O Box 2075ColomboSri Lanka

Location127 Sunil MawathaPelawattaBattaramullaSri Lanka

Telephone+94-11-2880000

Fax+94-11-2786854

[email protected]

Websitewww.iwmi.org

I n t e r n a t i o n a lWater ManagementI n s t i t u t e ISBN 978-92-9090-744-2

ISSN 1026-0862

Related Publications

Bartlett, R.; Bharati, L.; Pant, D.; Hosterman, H.; McCornick, P. G. 2010. Climate change impacts and adaptation in Nepal. Colombo, Sri Lanka: International Water Management Institute (IWMI). 35p. (IWMI Working Paper 139)

www.iwmi.org/Publications/Working_Papers/working/WOR139.pdf

Cai, X.; Sharma, B. R.; Matin, M. A.; Sharma, D.; Gunasinghe, S. 2010. An assessment of crop water productivity in the Indus and Ganges River Basins: current status and scope for improvement. Colombo, Sri Lanka: International Water Management Institute (IWMI). 30p. (IWMI Research Report 140)

www.iwmi.org/Publications/IWMI_Research_Reports/PDF/PUB140/RR140.pdf

Jha, R.; Smakhtin, V. 2008. A review of methods of hydrological estimation at ungauged sites in India. Colombo, Sri Lanka: International Water Management Institute (IWMI). 24p. (IWMI Working Paper 130)

www.iwmi.org/Publications/Working_Papers/working/WOR130.pdf

Smakhtin, V.; Gamage, M. S. D. N.; Bharati, L. 2007. Hydrological and environmental issues of interbasin water transfers in India: A case of the Krishna River Basin. Colombo, Sri Lanka: International Water Management Institute (IWMI). 35p. (IWMI Research Report 120)


Venot, J. P.; Turral, H.; Samad, M.; Molle, F. 2007. Shifting waterscapes: Explaining basin closure in the Lower Krishna Basin, South India. Colombo, Sri Lanka: International Water Management Institute (IWMI). 60p. (IWMI Research Report 121)


the impacts of water infrastructure and climate change on ......infrastructure and climate change on...

Documents